@arkstorm Unfortunately, a poll on SB would be subject to selection bias and recall bias and not give me objective unbiased data. I'm sure it would be interesting, however.
I respectfully disagree.
The selection bias would be offset by the factor that you are measuring failure per number of dives. In analyzing the data you would probably need to establish a minimum number of dives for data to be counted because, e.g., a very unlucky diver may have only had relatively few dives with a failure which would likely skew the results compared to the median number of dives per failure. But the sample group as a whole could be within reasonable tolerances.
Also, SPG/transmitter failure has no relation to the level of experience or enthusiasm of a diver, i.e., one that is more or less likely to be a scubaboard member, let alone one who would reply to a poll. I would agree with the contention that your average scubaboard member has more dives under the belt than the average diver but using respondents from this population is reasonable to represent scuba divers writ large when the data is per total number of dives as opposed to per diver.
Recall bias is always a problem in every poll but does not necessarily render it un-objective. An SPG/transmitter failure is likely not something a diver would either forget or falsely remember because what we know coming into the matter is that it does not happen often. Recall bias is managed by properly setting forth the poll question and defining the operative terms.
To maximize validity, I would pose the poll as follows:
1) How many dives have you completed in your lifetime (approximately)?
2) How many times has an SPG (tank pressure gauge connected by a hose) failed to provide you with your tank pressure after entering the water for a dive but before exiting the water?
3) How many times has a transmitter (tank pressure gauge connected wirelessly to a display device) failed to provide you with your tank pressure after entering the water for a dive but before exiting the water?
I would compile the data as such:
(a) I would eliminate responses from divers with less than 100 lifetime dives.
(b) I would apply a bell curve concept to define a median reported failure rate.
(c) I would define a margin of error based on the number of reports which fell outside the median.
(d) I would consider how many total reports (N=?) I would require to identify a statistically significant determination.
I believe that with enough responses the results of such a poll would be a valid indicator of failure rates over a large sample number of total dives.
I would not share my prediction/thesis prior to the poll to not bias the results.
Would be an interesting exercise if nothing else.
